{ai,lib,tools}[GCCcore/14.2.0] CTranslate2 v4.8.1, CCCL v2.7.0, cpu_features v0.10.1 w/ CUDA 12.8.0 by pavelToman · Pull Request #26603 · easybuilders/easybuild-easyconfigs

pavelToman · 2026-08-05T14:33:39Z

(created using eb --new-pr)
resolves vscentrum/vsc-software-stack#613

…-2.7.0-GCCcore-14.2.0-CUDA-12.8.0.eb, cpu_features-0.10.1-GCCcore-14.2.0.eb and patches: CTranslate2-4.8.1_fix-tests.patch, CTranslate2-4.8.1_use-system-dependencies.patch

github-actions · 2026-08-05T14:34:35Z

Updated software `CTranslate2-4.8.1-foss-2025a-CUDA-12.8.0.eb`

Diff against CTranslate2-4.5.0-foss-2023a-CUDA-12.6.0.eb

easybuild/easyconfigs/c/CTranslate2/CTranslate2-4.5.0-foss-2023a-CUDA-12.6.0.eb

diff --git a/easybuild/easyconfigs/c/CTranslate2/CTranslate2-4.5.0-foss-2023a-CUDA-12.6.0.eb b/easybuild/easyconfigs/c/CTranslate2/CTranslate2-4.8.1-foss-2025a-CUDA-12.8.0.eb
index a6376dba2e..984084a402 100644
--- a/easybuild/easyconfigs/c/CTranslate2/CTranslate2-4.5.0-foss-2023a-CUDA-12.6.0.eb
+++ b/easybuild/easyconfigs/c/CTranslate2/CTranslate2-4.8.1-foss-2025a-CUDA-12.8.0.eb
@@ -1,13 +1,13 @@
 easyblock = 'CMakeMake'
 
 name = 'CTranslate2'
-version = '4.5.0'
+version = '4.8.1'
 versionsuffix = '-CUDA-%(cudaver)s'
 
 homepage = 'https://opennmt.net/CTranslate2/'
 description = "Fast inference engine for Transformer models."
 
-toolchain = {'name': 'foss', 'version': '2023a'}
+toolchain = {'name': 'foss', 'version': '2025a'}
 
 source_urls = ['https://github.com/OpenNMT/CTranslate2/archive/']
 sources = [{
@@ -15,69 +15,81 @@ sources = [{
     "filename": SOURCE_TAR_GZ,
 }]
 patches = [
-    'CTranslate2-4.5.0_fix-third-party.patch',
-    'CTranslate2-4.5.0_fix-tests.patch',
-    'CTranslate2-4.5.0_replace-cxxopts.patch',
+    'CTranslate2-4.8.1_use-system-dependencies.patch',
+    'CTranslate2-4.8.1_fix-tests.patch',
 ]
 checksums = [
-    {'CTranslate2-4.5.0.tar.gz': 'f3040c7c3da5dde022fdc16906c279f3f936c6e79f3df8f998c908bb01a77cfe'},
-    {'CTranslate2-4.5.0_fix-third-party.patch': '45ab6d19954010dc5d515498a0827f0b13992d88b9691ab73ab27fee1114e3e3'},
-    {'CTranslate2-4.5.0_fix-tests.patch': '73123eafe612538354be5aa96c750199e1a219a5316800848c3894c1cc6ca2ad'},
-    {'CTranslate2-4.5.0_replace-cxxopts.patch': 'e378969c2968e2fd57863956a4d2f267731a49d1b890dcc45593d6a310531271'},
+    {'CTranslate2-4.8.1.tar.gz': '05c4e8a42c8573d0c79782e8408037c16db143334b6f4ee1cef23aa3b08a4dfa'},
+    {'CTranslate2-4.8.1_use-system-dependencies.patch':
+     '11efa9d05f9668bc21b09df58e022be0e4fbc974c80f3fcb099cb7df22c7b69d'},
+    {'CTranslate2-4.8.1_fix-tests.patch': 'b5b9c5997c7c00cb457752df74dc262b45e1b871ca9d89d48eecd213b68f9645'},
 ]
 
 builddependencies = [
-    ('CMake', '3.26.3'),
-    ('pybind11', '2.11.1'),
-    ('cxxopts', '3.0.0', '', SYSTEM),
-    ('spdlog', '1.11.0'),
-    ('cpu_features', '0.9.0'),
+    ('CMake', '3.31.3'),
+    ('pybind11', '2.13.6'),
+    ('spdlog', '1.15.3'),
+    ('cpu_features', '0.10.1'),
+    ('googletest', '1.17.0'),
 ]
-
 dependencies = [
-    ('CUDA', '12.6.0', '', SYSTEM),
-    ('Python', '3.11.3'),
-    ('SciPy-bundle', '2023.07'),
-    ('googletest', '1.13.0'),
-    ('PyYAML', '6.0'),
-    ('cuDNN', '9.5.0.50', versionsuffix, SYSTEM),
+    ('CUDA', '12.8.0', '', SYSTEM),
+    ('Python', '3.13.1'),
+    ('SciPy-bundle', '2025.06'),
+    ('PyYAML', '6.0.2'),
 ]
 
-# make sure that CTranslate2 libraries are linked to FlexiBLAS, not OpenBLAS
-preconfigopts = "export CMAKE_INCLUDE_PATH=$EBROOTFLEXIBLAS/include/flexiblas:${CMAKE_INCLUDE_PATH} && "
-preconfigopts += "sed -i 's/openblas/flexiblas/g' %(start_dir)s/CMakeLists.txt && "
-
-configopts = '-DOPENMP_RUNTIME=COMP -DWITH_CUDA=ON -DWITH_MKL=OFF '
-configopts += '-DOPENBLAS_INCLUDE_DIR="$EBROOTFLEXIBLAS/include" -DWITH_OPENBLAS=ON '
-configopts += '-DWITH_CUDNN=ON '
-configopts += '-DENABLE_CPU_DISPATCH=OFF '
-
-prebuildopts = 'export CT2_VERBOSE=3 && '
+configopts = ' '.join([
+    '-DWITH_CUDA=ON',
+    # Since v4.6.3, CTranslate2 has a native CUDA Conv1d implementation
+    '-DWITH_CUDNN=OFF',
+    '-DWITH_MKL=OFF',
+    '-DWITH_OPENBLAS=ON',
+    '-DOPENBLAS_INCLUDE_DIR="$EBROOTFLEXIBLAS/include/flexiblas"',
+    '-DOPENBLAS_LIBRARY="$EBROOTFLEXIBLAS/lib/libflexiblas.so"',
+    '-DOPENMP_RUNTIME=COMP',
+    # toolchain uses -march=native, which conflicts with CTranslate2's per-ISA runtime dispatch objects
+    '-DENABLE_CPU_DISPATCH=OFF',
+    # standalone C++ ct2-translator client is not required
+    '-DBUILD_CLI=OFF',
+    '-DBUILD_TESTS=ON',
+    '-DCUDA_ARCH_LIST="%(cuda_cc_space_sep)s"',
+])
 
 exts_defaultclass = 'PythonPackage'
 exts_default_options = {
-    'source_urls': [PYPI_SOURCE],
     'installopts': '',
     'runtest': False,
 }
 
 exts_list = [
     ('ctranslate2', version, {
+        'preinstallopts': "export CTRANSLATE2_ROOT=%(installdir)s && ",
         'sources': ['CTranslate2-%(version)s.tar.gz'],
         'start_dir': 'python',
-        'checksums': ['f3040c7c3da5dde022fdc16906c279f3f936c6e79f3df8f998c908bb01a77cfe'],
+        'checksums': ['05c4e8a42c8573d0c79782e8408037c16db143334b6f4ee1cef23aa3b08a4dfa'],
     }),
 ]
 
+test_cmd = './tests/ctranslate2_test %(start_dir)s/tests/data'
+runtest = True
+
 sanity_check_paths = {
-    'files': ['bin/ct2-translator', 'lib/libctranslate2.%s' % SHLIB_EXT],
-    'dirs': ['include/ctranslate2', 'lib/python%(pyshortver)s/site-packages'],
+    'files': [
+        'bin/ct2-transformers-converter',
+        'include/ctranslate2/translator.h',
+        'lib/libctranslate2.so',
+    ],
+    'dirs': [
+        'lib/cmake/ctranslate2',
+        'lib/python%(pyshortver)s/site-packages/ctranslate2',
+    ],
 }
 
 sanity_check_commands = [
-    "ct2-translator --help",
-    "python -c 'import ctranslate2'",
-    "python -m pip check",
+    "python -s -c 'import ctranslate2'",
+    "python -s -m pip check",
+    "ldd %(installdir)s/lib/libctranslate2.so | grep -q 'libflexiblas\\.so'",
 ]
 
 moduleclass = 'ai'

Updated software `cpu_features-0.10.1-GCCcore-14.2.0.eb`

Diff against cpu_features-0.9.0-GCCcore-12.3.0.eb

easybuild/easyconfigs/c/cpu_features/cpu_features-0.9.0-GCCcore-12.3.0.eb

diff --git a/easybuild/easyconfigs/c/cpu_features/cpu_features-0.9.0-GCCcore-12.3.0.eb b/easybuild/easyconfigs/c/cpu_features/cpu_features-0.10.1-GCCcore-14.2.0.eb
index 1d352eaf0a..9ea80dde05 100644
--- a/easybuild/easyconfigs/c/cpu_features/cpu_features-0.9.0-GCCcore-12.3.0.eb
+++ b/easybuild/easyconfigs/c/cpu_features/cpu_features-0.10.1-GCCcore-14.2.0.eb
@@ -1,23 +1,23 @@
 # This file is an EasyBuild reciPY as per https://github.com/easybuilders/easybuild
-# Author: Denis Kristak, update: Pavel Tománek
+# Author: Denis Kristak, update: Pavel Tomanek
 easyblock = 'CMakeMake'
 
 name = 'cpu_features'
-version = '0.9.0'
+version = '0.10.1'
 
 homepage = 'https://github.com/google/cpu_features'
 description = """A cross-platform C library to retrieve CPU features (such as available instructions) at runtime."""
 
-toolchain = {'name': 'GCCcore', 'version': '12.3.0'}
+toolchain = {'name': 'GCCcore', 'version': '14.2.0'}
 toolchainopts = {'pic': True}
 
 source_urls = ['https://github.com/google/cpu_features/archive/']
 sources = ['v%(version)s.tar.gz']
-checksums = ['bdb3484de8297c49b59955c3b22dba834401bc2df984ef5cfc17acbe69c5018e']
+checksums = ['52639b380fced11d738f8b151dbfee63fb94957731d07f1966c812e5b90cbad4']
 
 builddependencies = [
-    ('CMake', '3.26.3'),
-    ('binutils', '2.40'),
+    ('CMake', '3.31.3'),
+    ('binutils', '2.42'),
 ]
 
 modextrapaths = {MODULE_LOAD_ENV_HEADERS: 'include/cpu_features'}

pavelToman · 2026-08-05T14:35:18Z

@boegelbot please test @ jsc-zen3-a100

boegelbot · 2026-08-05T14:40:08Z

@pavelToman: Request for testing this PR well received on jsczen3l1.int.jsc-zen3.fz-juelich.de

PR test command 'if [[ develop != 'develop' ]]; then EB_BRANCH=develop ./easybuild_develop.sh 2> /dev/null 1>&2; EB_PREFIX=/home/boegelbot/easybuild/develop source init_env_easybuild_develop.sh; fi; EB_PR=26603 EB_ARGS= EB_CONTAINER= EB_REPO=easybuild-easyconfigs EB_BRANCH=develop /opt/software/slurm/bin/sbatch --job-name test_PR_26603 --ntasks=8 --partition=jsczen3g --gres=gpu:1 ~/boegelbot/eb_from_pr_upload_jsc-zen3.sh' executed!

exit code: 0
output:

Submitted batch job 11098

Test results coming soon (I hope)...

Details

- notification for comment with ID 5193147537 processed

Message to humans: this is just bookkeeping information for me,
it is of no use to you (unless you think I have a bug, which I don't).

pavelToman · 2026-08-05T14:44:46Z

Test report by @pavelToman
SUCCESS
Build succeeded for 3 out of 3 (total: 9 mins 21 secs) (3 easyconfigs in total)
node4004.donphan.os - Linux RHEL 9.6, x86_64, Intel(R) Xeon(R) Gold 6240 CPU @ 2.60GHz, 1 x NVIDIA NVIDIA A2, 595.71.05, Python 3.9.21
See https://gist.github.com/pavelToman/8d10be5ffbac92f7abec21bc667341cc for a full test report.

boegelbot · 2026-08-05T14:49:39Z

Test report by @boegelbot
SUCCESS
Build succeeded for 3 out of 3 (total: 9 mins 6 secs) (3 easyconfigs in total)
jsczen3g1.int.jsc-zen3.fz-juelich.de - Linux Rocky Linux 9.8, x86_64, AMD EPYC-Milan Processor (zen3), 1 x NVIDIA NVIDIA A100 80GB PCIe, 590.48.01, Python 3.9.25
See https://gist.github.com/boegelbot/6e3edaa5409dd0bc3c6b03f48dca6e65 for a full test report.

Thyre · 2026-08-06T06:48:17Z

I'm surprised that we still need CCCL. CUDA v12.8.0 should have everything included already.
Or are we using something that is missing from CUDA v12.8 but present in a newer CCCL version?

Edit: The CCCL version here matches the one bundled with CUDA, see https://docs.nvidia.com/cuda/archive/12.8.0/cuda-toolkit-release-notes/index.html#id4

pavelToman · 2026-08-06T08:44:39Z

I'm surprised that we still need CCCL. CUDA v12.8.0 should have everything included already. Or are we using something that is missing from CUDA v12.8 but present in a newer CCCL version?

Edit: The CCCL version here matches the one bundled with CUDA, see https://docs.nvidia.com/cuda/archive/12.8.0/cuda-toolkit-release-notes/index.html#id4

Thank you for the review. You are right, the CCCL is already included in CUDA module so no need to add EC for this.

….8.0.eb

pavelToman · 2026-08-06T08:45:48Z

@boegelbot please test @ jsc-zen3-a100

boegelbot · 2026-08-06T08:50:07Z

@pavelToman: Request for testing this PR well received on jsczen3l1.int.jsc-zen3.fz-juelich.de

PR test command 'if [[ develop != 'develop' ]]; then EB_BRANCH=develop ./easybuild_develop.sh 2> /dev/null 1>&2; EB_PREFIX=/home/boegelbot/easybuild/develop source init_env_easybuild_develop.sh; fi; EB_PR=26603 EB_ARGS= EB_CONTAINER= EB_REPO=easybuild-easyconfigs EB_BRANCH=develop /opt/software/slurm/bin/sbatch --job-name test_PR_26603 --ntasks=8 --partition=jsczen3g --gres=gpu:1 ~/boegelbot/eb_from_pr_upload_jsc-zen3.sh' executed!

exit code: 0
output:

Submitted batch job 11107

Test results coming soon (I hope)...

Details

- notification for comment with ID 5202423423 processed

Message to humans: this is just bookkeeping information for me,
it is of no use to you (unless you think I have a bug, which I don't).

pavelToman · 2026-08-06T08:52:37Z

Test report by @pavelToman
SUCCESS
Build succeeded for 2 out of 2 (total: 6 mins 22 secs) (2 easyconfigs in total)
node4014.donphan.os - Linux RHEL 9.6, x86_64, Intel(R) Xeon(R) Gold 6240 CPU @ 2.60GHz, 1 x NVIDIA NVIDIA A2, 595.71.05, Python 3.9.21
See https://gist.github.com/pavelToman/3a5403757fdbf4f78b182e5dcab58d4b for a full test report.

boegelbot · 2026-08-06T08:58:38Z

Test report by @boegelbot
SUCCESS
Build succeeded for 2 out of 2 (total: 6 mins 4 secs) (2 easyconfigs in total)
jsczen3g1.int.jsc-zen3.fz-juelich.de - Linux Rocky Linux 9.8, x86_64, AMD EPYC-Milan Processor (zen3), 1 x NVIDIA NVIDIA A100 80GB PCIe, 590.48.01, Python 3.9.25
See https://gist.github.com/boegelbot/5595d2c69e43a4926fbaf330361a4f2e for a full test report.

adding easyconfigs: CTranslate2-4.8.1-foss-2025a-CUDA-12.8.0.eb, CCCL…

040733f

…-2.7.0-GCCcore-14.2.0-CUDA-12.8.0.eb, cpu_features-0.10.1-GCCcore-14.2.0.eb and patches: CTranslate2-4.8.1_fix-tests.patch, CTranslate2-4.8.1_use-system-dependencies.patch

pavelToman added the update label Aug 5, 2026

pavelToman mentioned this pull request Aug 5, 2026

whisperx vscentrum/vsc-software-stack#613

Open

2 tasks

github-actions Bot added the 2025a issues & PRs related to 2025a common toolchains label Aug 5, 2026

pavelToman added the AI-assisted AI-assisted contributions label Aug 5, 2026

remove CCCL as dep

a4c5f8b

Delete easybuild/easyconfigs/c/CCCL/CCCL-2.7.0-GCCcore-14.2.0-CUDA-12…

7e2a663

….8.0.eb

Navigation Menu

Uh oh!

{ai,lib,tools}[GCCcore/14.2.0] CTranslate2 v4.8.1, CCCL v2.7.0, cpu_features v0.10.1 w/ CUDA 12.8.0 - #26603

{ai,lib,tools}[GCCcore/14.2.0] CTranslate2 v4.8.1, CCCL v2.7.0, cpu_features v0.10.1 w/ CUDA 12.8.0#26603
pavelToman wants to merge 3 commits into
easybuilders:developfrom
pavelToman:20260805163335_new_pr_CTranslate2481

pavelToman commented Aug 5, 2026 •

edited

Loading

Uh oh!

github-actions Bot commented Aug 5, 2026 •

edited

Loading

Uh oh!

pavelToman commented Aug 5, 2026

Uh oh!

boegelbot commented Aug 5, 2026

Uh oh!

pavelToman commented Aug 5, 2026

Uh oh!

boegelbot commented Aug 5, 2026

Uh oh!

Thyre commented Aug 6, 2026 •

edited

Loading

Uh oh!

pavelToman commented Aug 6, 2026 •

edited

Loading

Uh oh!

pavelToman commented Aug 6, 2026

Uh oh!

boegelbot commented Aug 6, 2026

Uh oh!

pavelToman commented Aug 6, 2026

Uh oh!

boegelbot commented Aug 6, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

Uh oh!

Conversation

pavelToman commented Aug 5, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

github-actions Bot commented Aug 5, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Updated software CTranslate2-4.8.1-foss-2025a-CUDA-12.8.0.eb

Updated software cpu_features-0.10.1-GCCcore-14.2.0.eb

Uh oh!

pavelToman commented Aug 5, 2026

Uh oh!

boegelbot commented Aug 5, 2026

Uh oh!

pavelToman commented Aug 5, 2026

Uh oh!

boegelbot commented Aug 5, 2026

Uh oh!

Thyre commented Aug 6, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

pavelToman commented Aug 6, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

pavelToman commented Aug 6, 2026

Uh oh!

boegelbot commented Aug 6, 2026

Uh oh!

pavelToman commented Aug 6, 2026

Uh oh!

boegelbot commented Aug 6, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

pavelToman commented Aug 5, 2026 •

edited

Loading

github-actions Bot commented Aug 5, 2026 •

edited

Loading

Updated software `CTranslate2-4.8.1-foss-2025a-CUDA-12.8.0.eb`

Updated software `cpu_features-0.10.1-GCCcore-14.2.0.eb`

Thyre commented Aug 6, 2026 •

edited

Loading

pavelToman commented Aug 6, 2026 •

edited

Loading